Predictika · Research model orders

An AI sales agent's research animal order, checked before it goes out

An AI sales agent prepares an order for research mice or rats: the strain, the type of order, the weight range, the pregnancy terms, the number of animals, the crates, the shipping zone, the packing and the total the quote states. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of how these choices fit together, and returns a verdict, every reason and the total it computed.

Scope: One research model order: strain, order type, weight range, pregnancy terms, crates, shipping, packing and the stated total. See everything this validator covers

This validator models research model orders for seven strains, not just the scenarios shown below.

  • Strains and order types — CD-1 IGS, C57BL/6 and BALB/c mice, CD IGS, Wistar Han IGS and Lewis rats and nude mice, ordered as standard, timed-pregnant or retired-breeder stock.
  • Weight ranges and surcharges — which weight ranges each kind of stock takes, the surcharge each range carries and the surcharge for retired breeders with a specified age or weight.
  • Pregnancy terms — the gestation days a timed-pregnant order can name, the pregnancy guarantee for outbred and inbred stock, and the plug-only guarantee before gestation day 13.
  • Crates, shipping and packing — how many animals a crate holds, the shipping zone, the laminar-flow packing and barrier housing that immunodeficient animals need, the size of the order and the total stated on the quote, which is compared with the computed one.

Any research model order for these strains that touches these rules is in scope — not only the scenarios pictured here. Not covered: other strains and services, and safety-assessment studies. The strain names, the weight-range and retired-breeder surcharges, the pregnancy guarantees and the packing for immunodeficient animals follow Charles River's public ordering information; the unit prices, crate capacities, shipping zones and order limits are modeled from comparable products and are not published by Charles River.

The AI sales agentBuilds the order from the conversation with the researcher and says it is ready to send.
Predictika validatorChecks every rule at once and returns the verdict, each finding, and the computed figures.

7 scenarios in 3 groups · each group runs from the agent’s first attempt to the outcome · click one to switch

What the agent proposes

The AI sales agent says

Illustrative agent statement. The validator receives the structured request below.

Proposing agent: AI sales agent (scripted)

Before sending, the agent must call Recorded run

Predictika validator


      
Waiting for the call.

Where rules-based CPQ gets it wrong

Click to expand

Our assessment of what a typical rules-based CPQ (product rules, price rules and validation rules, as in Salesforce CPQ) does with the same cases. It is a judgment about effort and reliability, not a claim that nothing can be built.

What the validator caughtTypical rules-based CPQPredictika
Pregnancy terms that depend on the stockOnly with care
The guarantee is 100 percent for outbred stock and 75 percent for inbred stock from gestation day 13, and only the plug before that, so each promise needs its own rule on the stock and the day.
The stock, the gestation day and the promised percentage are fields of one order and are checked together on every call.
Weight ranges that only some stock takesOnly with care
Inbred mice take a 1 g range and rats take 5 g or 10 g, so every range needs a rule on the kind of stock.
The weight range is checked against the strain in the same pass as everything else.
Packing and housing for immunodeficient animalsOnly with workarounds
Nude mice cannot be timed pregnant, need laminar-flow packing and need barrier housing at the destination, three rules that each depend on the strain.
The strain sets all three requirements at once, and each one is reported separately.
A stated total that must match the computed oneOnly with workarounds
The price adds the animals with their surcharge, the pregnancy fee and the shipping per crate, so a separate rule must recompute it and compare it with the stated figure.
The stated total is a field of the order and is compared with the computed total on every call.
Every problem returned in one answerOnly with workarounds
Rules tend to fire as a step is completed, so a rep often fixes one error and meets the next.
All findings come back together, so the agent can fix them in a single pass.

What this validator catches: findings from its test cases

Click to expand

Every rule is checked on every call, in any order the choices arrive; one problem never hides another. Blocking findings stop the order; approval findings need a person to sign off; everything else is computed.

1A weight range of 1 g is not offered for this stock; inbred mice take 1 g and rats take 5 g or 10 g.
2The order promises 100 percent pregnant at shipment; this stock is guaranteed 75 percent at gestation day 13 or later.
3The order promises 100 percent pregnant, but up to gestation day 12 only the plug is guaranteed.
4A timed-pregnant order needs a gestation day from 1 to 19; the order has 21.
5Timed-pregnant orders are not offered for immunodeficient stock.
6Immunodeficient animals are packed under laminar flow hoods; the order does not request it.
7Immunodeficient animals need barrier housing at the destination.
84 crates hold 40 animals; the order has 41.
9An order of 2,000 animals needs a production capacity review.
10The order has 2,001 animals; one order allows at most 2,000.
11A zone 3 shipment needs a carrier and transit-time check.
12The quote states a total of $1,000; the order items give $3,120.

Scenarios and results

Click to expand

The recorded result of the validator for each scenario on this page: the request the AI agent sent, the verdict, and every finding, in the order they appear above.

1. Nude mice: Timed-pregnant nude mice in one crate, shipped to zone 3 for a destination without barrier housing

The agent's request. Please quote an order of 30 timed-pregnant nude mice at gestation day 15, in 1 crate, shipped to zone 3, with standard packing, to a destination without barrier housing. The quote states $3,860. It is ready to send.

Verdict: Blocked.

  • Immunodeficient animals not timed pregnant. Timed-pregnant orders are not offered for immunodeficient stock.
  • Not enough crates. 1 crates hold 20 animals; the order has 30.
  • Immunodeficient animals need barrier housing. Immunodeficient animals need barrier housing at the destination.
  • Immunodeficient animals need laminar-flow packing. Immunodeficient animals are packed under laminar flow hoods; the order does not request it.
  • Long-distance shipment. A zone 3 shipment needs a carrier and transit-time check.

2. Nude mice: The agent resubmits after fixing the errors

The agent's request. Please quote an order of 30 standard nude mice, in 2 crates, shipped to zone 3, packed under laminar flow hoods, to a destination without barrier housing. The quote states $3,520. It is ready to send.

Verdict: Blocked.

  • Immunodeficient animals need barrier housing. Immunodeficient animals need barrier housing at the destination.
  • Long-distance shipment. A zone 3 shipment needs a carrier and transit-time check.

3. Nude mice: The agent resubmits after fixing the errors

The agent's request. Please quote an order of 30 standard nude mice, in 2 crates, shipped to zone 2, packed under laminar flow hoods, to a barrier housing destination. The quote states $3,360. It is ready to send.

Verdict: Allowed.

  • Order summary. 30 animals in 2 crate(s); animals $3,000, pregnancy fee $0, shipping $360; total $3,360.

4. Pregnant rats: A large order of timed-pregnant Lewis rats with a promise and a weight range the stock does not take

The agent's request. Please quote an order of 450 timed-pregnant Lewis rats at gestation day 10, promising 100 percent pregnant at shipment, with a 7 g weight range, in 45 crates, shipped to zone 1. The quote states $50,400. It is ready to send.

Verdict: Blocked.

  • Weight range not available. A weight range of 7 g is not offered for this stock; inbred mice take 1 g and rats take 5 g or 10 g.
  • Guarantee before day 13 is plug only. The order promises 100 percent pregnant, but up to gestation day 12 only the plug is guaranteed.
  • Large order. An order of 450 animals needs a production capacity review.

5. Pregnant rats: The agent resubmits after fixing the errors

The agent's request. Please quote an order of 400 timed-pregnant Lewis rats at gestation day 15, promising 75 percent pregnant at shipment, with a 10 g weight range, in 40 crates, shipped to zone 1. The quote states $49,600. It is ready to send.

Verdict: Allowed.

  • Order summary. 400 animals in 40 crate(s); animals $36,800, pregnancy fee $8,000, shipping $4,800; total $49,600.

6. Mice with a wrong total: Sixty CD-1 mice in two crates, with a weight range and a total that does not match

The agent's request. Please quote an order of 60 standard CD-1 IGS mice, with a 1 g weight range, in 2 crates, shipped to zone 2. The quote states $1,000. It is ready to send.

Verdict: Blocked.

  • Weight range not available. A weight range of 1 g is not offered for this stock; inbred mice take 1 g and rats take 5 g or 10 g.
  • Not enough crates. 2 crates hold 50 animals; the order has 60.
  • Stated total does not match. The quote states a total of $1,000; the order items give $1,560.

7. Mice with a wrong total: The agent resubmits after fixing the errors

The agent's request. Please quote an order of 60 standard CD-1 IGS mice, in 3 crates, shipped to zone 2. The quote states $1,740. It is ready to send.

Verdict: Allowed.

  • Order summary. 60 animals in 3 crate(s); animals $1,200, pregnancy fee $0, shipping $540; total $1,740.

Narration transcript

Click to expand

The text of the narrated walkthrough, step by step.

  1. Welcome to a demo of Predictika's Validator. An AI sales agent prepares an order for research mice or rats: the strain, the type of order, the weight range, and more. Before it goes out, the agent sends it to Predictika, which checks it against a declarative model of how these choices fit together, and returns a verdict, every reason and the total it computed.
  2. I will start with the first request by the agent.
  3. The agent says. Please quote an order of 30 timed-pregnant nude mice at gestation day 15, in 1 crate, shipped to zone 3, with standard packing, to a destination without barrier housing. The quote states $3,860. The rest of the request is on screen.
  4. Before it goes out, the agent calls the validator.
  5. Blocked, with four findings and one approval, all in one answer. Timed-pregnant orders are not offered for immunodeficient stock. 1 crates hold 20 animals; the order has 30. Immunodeficient animals need barrier housing at the destination. The other one is on screen.
  6. The validator returned four blocking findings and one approval in one answer. A check that reads one field at a time would pass most of this; it shows only when the choices in the order are checked against each other.
  7. What this shows: for packing and housing for immunodeficient animals, a rules-based CPQ can do this only with workarounds.
  8. Now the agent resubmits.
  9. The corrected request. It changes the order type to standard, the gestation day to not applicable, the crates to 2, the laminar-flow packing to yes and the total stated on the quote to $3,520.
  10. It calls the validator again.
  11. Still blocked, with one finding and one approval, all in one answer. Immunodeficient animals need barrier housing at the destination.
  12. Three findings cleared. Still blocking: immunodeficient animals need barrier housing. One approval also remains for a person: long-distance shipment.
  13. The agent resubmits once more.
  14. The revised request. It changes the shipping zone to 2, the barrier housing at destination to yes and the total stated on the quote to $3,360.
  15. It calls the validator once more.
  16. This time the validator clears it.
  17. Both earlier findings are cleared.
  18. Five problems caught in one call, fixed in two passes, and a clean $3,360 order. That is the value of Predictika's validator: catching mistakes before they reach the customer. Now you can try the other scenarios yourself.